适合于生物图像的图像融合算法研究

图像融合作为一种有效的信息融合的技术，已广泛用于军事、遥感、机器视觉和医学图像等领域。本文讨论了三种基于像素级的图像融合算法：Ｐ加权平均，Ｔｏｅｔ算法和基于小波变换的算法；采用四种评价融合效果的量化判据：标准偏差，平均误差，峰值信噪比（ｓ／Ｎ）ｐ和熵差，将三种图像融合算法用于生物图像中细胞荧光图像和透射图像的融合，量化评价结果和视觉判断均说明，对于以细胞荧光图像和透射图像研究对象，需突出荧光图像的

Study of Image Fusion Methods Appropriated to Biological Images

Being an available method of information fusion, image fusion has been used in many fields such as military applications, remote sensing, machine vision and medical images. Three image fusion algorithms are presented based on pixel level including weighted mean of the original images, Toet algorithm, and the algorithm based on wavelet transform. Four kinds of quantitative evaluation criteria for the quality of image fusion algorithms are proposed such as standard deviation, average error, peak signal noise rate and the difference of entropy between the ideal image and the fusion image. Using the four algorithms to merge the fluorescence image and transmission image, the conclusions of the quantitative evaluation and the human vision are similar.